natural 0.2.0 → 0.2.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.editorconfig +7 -0
- package/README.md +3 -3
- package/lib/natural/inflectors/present_verb_inflector.js +1 -0
- package/lib/natural/phonetics/double_metaphone.js +9 -4
- package/lib/natural/phonetics/metaphone.js +4 -0
- package/lib/natural/phonetics/soundex.js +11 -7
- package/lib/natural/stemmers/porter_stemmer.js +90 -62
- package/lib/natural/stemmers/porter_stemmer_pt.js +302 -0
- package/lib/natural/stemmers/stemmer_pt.js +65 -0
- package/lib/natural/stemmers/token.js +163 -0
- package/lib/natural/util/stopwords_pt.js +136 -0
- package/package.json +2 -2
- package/spec/double_metaphone_spec.js +10 -1
- package/spec/metaphone_spec.js +5 -0
- package/spec/porter_stemmer_pt_spec.js +50 -0
- package/spec/porter_stemmer_spec.js +45 -0
- package/spec/present_verb_inflector_spec.js +3 -0
- package/spec/soundex_spec.js +13 -9
- package/spec/stemmer_token_spec.js +161 -0
- package/spec/test_data/snowball_pt.txt +32016 -0
- package/coverage/coverage.json +0 -1
- package/coverage/lcov-report/index.html +0 -584
- package/coverage/lcov-report/io_spec/index.html +0 -350
- package/coverage/lcov-report/io_spec/tfidf_spec.js.html +0 -469
- package/coverage/lcov-report/lib/natural/analyzers/index.html +0 -350
- package/coverage/lcov-report/lib/natural/analyzers/sentence_analyzer.js.html +0 -850
- package/coverage/lcov-report/lib/natural/classifiers/bayes_classifier.js.html +0 -496
- package/coverage/lcov-report/lib/natural/classifiers/classifier.js.html +0 -823
- package/coverage/lcov-report/lib/natural/classifiers/index.html +0 -376
- package/coverage/lcov-report/lib/natural/classifiers/logistic_regression_classifier.js.html +0 -502
- package/coverage/lcov-report/lib/natural/distance/dice_coefficient.js.html +0 -529
- package/coverage/lcov-report/lib/natural/distance/index.html +0 -376
- package/coverage/lcov-report/lib/natural/distance/jaro-winkler_distance.js.html +0 -670
- package/coverage/lcov-report/lib/natural/distance/levenshtein_distance.js.html +0 -526
- package/coverage/lcov-report/lib/natural/index.html +0 -350
- package/coverage/lcov-report/lib/natural/index.js.html +0 -547
- package/coverage/lcov-report/lib/natural/inflectors/count_inflector.js.html +0 -484
- package/coverage/lcov-report/lib/natural/inflectors/form_set.js.html +0 -406
- package/coverage/lcov-report/lib/natural/inflectors/fr/count_inflector.js.html +0 -520
- package/coverage/lcov-report/lib/natural/inflectors/fr/index.html +0 -363
- package/coverage/lcov-report/lib/natural/inflectors/fr/noun_inflector.js.html +0 -1027
- package/coverage/lcov-report/lib/natural/inflectors/index.html +0 -402
- package/coverage/lcov-report/lib/natural/inflectors/ja/index.html +0 -350
- package/coverage/lcov-report/lib/natural/inflectors/ja/noun_inflector.js.html +0 -727
- package/coverage/lcov-report/lib/natural/inflectors/noun_inflector.js.html +0 -637
- package/coverage/lcov-report/lib/natural/inflectors/present_verb_inflector.js.html +0 -544
- package/coverage/lcov-report/lib/natural/inflectors/singular_plural_inflector.js.html +0 -649
- package/coverage/lcov-report/lib/natural/ngrams/index.html +0 -363
- package/coverage/lcov-report/lib/natural/ngrams/ngrams.js.html +0 -595
- package/coverage/lcov-report/lib/natural/ngrams/ngrams_zh.js.html +0 -571
- package/coverage/lcov-report/lib/natural/normalizers/index.html +0 -389
- package/coverage/lcov-report/lib/natural/normalizers/normalizer.js.html +0 -607
- package/coverage/lcov-report/lib/natural/normalizers/normalizer_ja.js.html +0 -2194
- package/coverage/lcov-report/lib/natural/normalizers/normalizer_no.js.html +0 -496
- package/coverage/lcov-report/lib/natural/normalizers/remove_diacritics.js.html +0 -685
- package/coverage/lcov-report/lib/natural/phonetics/dm_soundex.js.html +0 -1063
- package/coverage/lcov-report/lib/natural/phonetics/double_metaphone.js.html +0 -1840
- package/coverage/lcov-report/lib/natural/phonetics/index.html +0 -402
- package/coverage/lcov-report/lib/natural/phonetics/metaphone.js.html +0 -886
- package/coverage/lcov-report/lib/natural/phonetics/phonetic.js.html +0 -484
- package/coverage/lcov-report/lib/natural/phonetics/soundex.js.html +0 -562
- package/coverage/lcov-report/lib/natural/spellcheck/index.html +0 -350
- package/coverage/lcov-report/lib/natural/spellcheck/spellcheck.js.html +0 -562
- package/coverage/lcov-report/lib/natural/stemmers/index.html +0 -571
- package/coverage/lcov-report/lib/natural/stemmers/lancaster_rules.js.html +0 -2692
- package/coverage/lcov-report/lib/natural/stemmers/lancaster_stemmer.js.html +0 -544
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer.js.html +0 -937
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_es.js.html +0 -985
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_fa.js.html +0 -418
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_it.js.html +0 -1018
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_no.js.html +0 -784
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_ru.js.html +0 -775
- package/coverage/lcov-report/lib/natural/stemmers/stemmer.js.html +0 -505
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_es.js.html +0 -496
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_fa.js.html +0 -484
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_fr.js.html +0 -496
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_it.js.html +0 -427
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_ja.js.html +0 -736
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_no.js.html +0 -505
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_pl.js.html +0 -496
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_ru.js.html +0 -496
- package/coverage/lcov-report/lib/natural/tfidf/index.html +0 -350
- package/coverage/lcov-report/lib/natural/tfidf/tfidf.js.html +0 -904
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
- package/coverage/lcov-report/lib/natural/tokenizers/index.html +0 -519
- package/coverage/lcov-report/lib/natural/tokenizers/regexp_tokenizer.js.html +0 -580
- package/coverage/lcov-report/lib/natural/tokenizers/tokenizer.js.html +0 -475
- package/coverage/lcov-report/lib/natural/tokenizers/tokenizer_ja.js.html +0 -1153
- package/coverage/lcov-report/lib/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
- package/coverage/lcov-report/lib/natural/transliterators/ja/index.html +0 -350
- package/coverage/lcov-report/lib/natural/transliterators/ja/index.js.html +0 -2068
- package/coverage/lcov-report/lib/natural/trie/index.html +0 -350
- package/coverage/lcov-report/lib/natural/trie/trie.js.html +0 -1012
- package/coverage/lcov-report/lib/natural/util/bag.js.html +0 -472
- package/coverage/lcov-report/lib/natural/util/edge_weighted_digraph.js.html +0 -658
- package/coverage/lcov-report/lib/natural/util/index.html +0 -532
- package/coverage/lcov-report/lib/natural/util/longest_path_tree.js.html +0 -619
- package/coverage/lcov-report/lib/natural/util/shortest_path_tree.js.html +0 -622
- package/coverage/lcov-report/lib/natural/util/stopwords.js.html +0 -445
- package/coverage/lcov-report/lib/natural/util/stopwords_es.js.html +0 -430
- package/coverage/lcov-report/lib/natural/util/stopwords_fa.js.html +0 -436
- package/coverage/lcov-report/lib/natural/util/stopwords_fr.js.html +0 -904
- package/coverage/lcov-report/lib/natural/util/stopwords_it.js.html +0 -478
- package/coverage/lcov-report/lib/natural/util/stopwords_ja.js.html +0 -499
- package/coverage/lcov-report/lib/natural/util/stopwords_no.js.html +0 -445
- package/coverage/lcov-report/lib/natural/util/stopwords_pl.js.html +0 -508
- package/coverage/lcov-report/lib/natural/util/stopwords_ru.js.html +0 -445
- package/coverage/lcov-report/lib/natural/util/topological.js.html +0 -616
- package/coverage/lcov-report/lib/natural/util/utils.js.html +0 -679
- package/coverage/lcov-report/lib/natural/wordnet/data_file.js.html +0 -580
- package/coverage/lcov-report/lib/natural/wordnet/index.html +0 -389
- package/coverage/lcov-report/lib/natural/wordnet/index_file.js.html +0 -742
- package/coverage/lcov-report/lib/natural/wordnet/wordnet.js.html +0 -829
- package/coverage/lcov-report/lib/natural/wordnet/wordnet_file.js.html +0 -529
- package/coverage/lcov-report/natural/analyzers/index.html +0 -350
- package/coverage/lcov-report/natural/analyzers/sentence_analyzer.js.html +0 -850
- package/coverage/lcov-report/natural/classifiers/bayes_classifier.js.html +0 -496
- package/coverage/lcov-report/natural/classifiers/classifier.js.html +0 -823
- package/coverage/lcov-report/natural/classifiers/index.html +0 -376
- package/coverage/lcov-report/natural/classifiers/logistic_regression_classifier.js.html +0 -502
- package/coverage/lcov-report/natural/distance/dice_coefficient.js.html +0 -529
- package/coverage/lcov-report/natural/distance/index.html +0 -376
- package/coverage/lcov-report/natural/distance/jaro-winkler_distance.js.html +0 -670
- package/coverage/lcov-report/natural/distance/levenshtein_distance.js.html +0 -526
- package/coverage/lcov-report/natural/index.html +0 -350
- package/coverage/lcov-report/natural/index.js.html +0 -433
- package/coverage/lcov-report/natural/inflectors/count_inflector.js.html +0 -484
- package/coverage/lcov-report/natural/inflectors/form_set.js.html +0 -406
- package/coverage/lcov-report/natural/inflectors/fr/index.html +0 -350
- package/coverage/lcov-report/natural/inflectors/fr/noun_inflector.js.html +0 -1027
- package/coverage/lcov-report/natural/inflectors/index.html +0 -402
- package/coverage/lcov-report/natural/inflectors/ja/index.html +0 -350
- package/coverage/lcov-report/natural/inflectors/ja/noun_inflector.js.html +0 -727
- package/coverage/lcov-report/natural/inflectors/noun_inflector.js.html +0 -637
- package/coverage/lcov-report/natural/inflectors/present_verb_inflector.js.html +0 -544
- package/coverage/lcov-report/natural/inflectors/singular_plural_inflector.js.html +0 -649
- package/coverage/lcov-report/natural/lib/natural/analyzers/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/analyzers/sentence_analyzer.js.html +0 -850
- package/coverage/lcov-report/natural/lib/natural/classifiers/bayes_classifier.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/classifiers/classifier.js.html +0 -823
- package/coverage/lcov-report/natural/lib/natural/classifiers/index.html +0 -376
- package/coverage/lcov-report/natural/lib/natural/classifiers/logistic_regression_classifier.js.html +0 -502
- package/coverage/lcov-report/natural/lib/natural/distance/dice_coefficient.js.html +0 -529
- package/coverage/lcov-report/natural/lib/natural/distance/index.html +0 -376
- package/coverage/lcov-report/natural/lib/natural/distance/jaro-winkler_distance.js.html +0 -670
- package/coverage/lcov-report/natural/lib/natural/distance/levenshtein_distance.js.html +0 -526
- package/coverage/lcov-report/natural/lib/natural/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/index.js.html +0 -547
- package/coverage/lcov-report/natural/lib/natural/inflectors/count_inflector.js.html +0 -484
- package/coverage/lcov-report/natural/lib/natural/inflectors/form_set.js.html +0 -406
- package/coverage/lcov-report/natural/lib/natural/inflectors/fr/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/inflectors/fr/noun_inflector.js.html +0 -1027
- package/coverage/lcov-report/natural/lib/natural/inflectors/index.html +0 -402
- package/coverage/lcov-report/natural/lib/natural/inflectors/ja/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/inflectors/ja/noun_inflector.js.html +0 -727
- package/coverage/lcov-report/natural/lib/natural/inflectors/noun_inflector.js.html +0 -637
- package/coverage/lcov-report/natural/lib/natural/inflectors/present_verb_inflector.js.html +0 -544
- package/coverage/lcov-report/natural/lib/natural/inflectors/singular_plural_inflector.js.html +0 -649
- package/coverage/lcov-report/natural/lib/natural/ngrams/index.html +0 -363
- package/coverage/lcov-report/natural/lib/natural/ngrams/ngrams.js.html +0 -595
- package/coverage/lcov-report/natural/lib/natural/ngrams/ngrams_zh.js.html +0 -571
- package/coverage/lcov-report/natural/lib/natural/normalizers/index.html +0 -389
- package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer.js.html +0 -607
- package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer_ja.js.html +0 -2194
- package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer_no.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/normalizers/remove_diacritics.js.html +0 -685
- package/coverage/lcov-report/natural/lib/natural/phonetics/dm_soundex.js.html +0 -1063
- package/coverage/lcov-report/natural/lib/natural/phonetics/double_metaphone.js.html +0 -1840
- package/coverage/lcov-report/natural/lib/natural/phonetics/index.html +0 -402
- package/coverage/lcov-report/natural/lib/natural/phonetics/metaphone.js.html +0 -886
- package/coverage/lcov-report/natural/lib/natural/phonetics/phonetic.js.html +0 -484
- package/coverage/lcov-report/natural/lib/natural/phonetics/soundex.js.html +0 -562
- package/coverage/lcov-report/natural/lib/natural/stemmers/index.html +0 -571
- package/coverage/lcov-report/natural/lib/natural/stemmers/lancaster_rules.js.html +0 -2692
- package/coverage/lcov-report/natural/lib/natural/stemmers/lancaster_stemmer.js.html +0 -544
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer.js.html +0 -937
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_es.js.html +0 -985
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_fa.js.html +0 -418
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_it.js.html +0 -1018
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_no.js.html +0 -784
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_ru.js.html +0 -775
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer.js.html +0 -505
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_es.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_fa.js.html +0 -484
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_fr.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_it.js.html +0 -427
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_ja.js.html +0 -736
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_no.js.html +0 -505
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_pl.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_ru.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/tfidf/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/tfidf/tfidf.js.html +0 -904
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
- package/coverage/lcov-report/natural/lib/natural/tokenizers/index.html +0 -519
- package/coverage/lcov-report/natural/lib/natural/tokenizers/regexp_tokenizer.js.html +0 -580
- package/coverage/lcov-report/natural/lib/natural/tokenizers/tokenizer.js.html +0 -475
- package/coverage/lcov-report/natural/lib/natural/tokenizers/tokenizer_ja.js.html +0 -1153
- package/coverage/lcov-report/natural/lib/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
- package/coverage/lcov-report/natural/lib/natural/transliterators/ja/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/transliterators/ja/index.js.html +0 -2068
- package/coverage/lcov-report/natural/lib/natural/trie/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/trie/trie.js.html +0 -1012
- package/coverage/lcov-report/natural/lib/natural/util/bag.js.html +0 -472
- package/coverage/lcov-report/natural/lib/natural/util/edge_weighted_digraph.js.html +0 -658
- package/coverage/lcov-report/natural/lib/natural/util/index.html +0 -532
- package/coverage/lcov-report/natural/lib/natural/util/longest_path_tree.js.html +0 -619
- package/coverage/lcov-report/natural/lib/natural/util/shortest_path_tree.js.html +0 -622
- package/coverage/lcov-report/natural/lib/natural/util/stopwords.js.html +0 -445
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_es.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_fa.js.html +0 -436
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_fr.js.html +0 -904
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_it.js.html +0 -478
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_ja.js.html +0 -499
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_no.js.html +0 -445
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_pl.js.html +0 -508
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_ru.js.html +0 -445
- package/coverage/lcov-report/natural/lib/natural/util/topological.js.html +0 -616
- package/coverage/lcov-report/natural/lib/natural/util/utils.js.html +0 -679
- package/coverage/lcov-report/natural/lib/natural/wordnet/data_file.js.html +0 -580
- package/coverage/lcov-report/natural/lib/natural/wordnet/index.html +0 -389
- package/coverage/lcov-report/natural/lib/natural/wordnet/index_file.js.html +0 -742
- package/coverage/lcov-report/natural/lib/natural/wordnet/wordnet.js.html +0 -829
- package/coverage/lcov-report/natural/lib/natural/wordnet/wordnet_file.js.html +0 -529
- package/coverage/lcov-report/natural/ngrams/index.html +0 -363
- package/coverage/lcov-report/natural/ngrams/ngrams.js.html +0 -595
- package/coverage/lcov-report/natural/ngrams/ngrams_zh.js.html +0 -571
- package/coverage/lcov-report/natural/normalizers/index.html +0 -389
- package/coverage/lcov-report/natural/normalizers/normalizer.js.html +0 -607
- package/coverage/lcov-report/natural/normalizers/normalizer_ja.js.html +0 -2194
- package/coverage/lcov-report/natural/normalizers/normalizer_no.js.html +0 -496
- package/coverage/lcov-report/natural/normalizers/remove_diacritics.js.html +0 -685
- package/coverage/lcov-report/natural/phonetics/dm_soundex.js.html +0 -1063
- package/coverage/lcov-report/natural/phonetics/double_metaphone.js.html +0 -1840
- package/coverage/lcov-report/natural/phonetics/index.html +0 -402
- package/coverage/lcov-report/natural/phonetics/metaphone.js.html +0 -886
- package/coverage/lcov-report/natural/phonetics/phonetic.js.html +0 -484
- package/coverage/lcov-report/natural/phonetics/soundex.js.html +0 -562
- package/coverage/lcov-report/natural/stemmers/index.html +0 -571
- package/coverage/lcov-report/natural/stemmers/lancaster_rules.js.html +0 -2692
- package/coverage/lcov-report/natural/stemmers/lancaster_stemmer.js.html +0 -544
- package/coverage/lcov-report/natural/stemmers/porter_stemmer.js.html +0 -937
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_es.js.html +0 -985
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_fa.js.html +0 -418
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_it.js.html +0 -1018
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_no.js.html +0 -784
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_ru.js.html +0 -775
- package/coverage/lcov-report/natural/stemmers/stemmer.js.html +0 -505
- package/coverage/lcov-report/natural/stemmers/stemmer_es.js.html +0 -496
- package/coverage/lcov-report/natural/stemmers/stemmer_fa.js.html +0 -484
- package/coverage/lcov-report/natural/stemmers/stemmer_fr.js.html +0 -496
- package/coverage/lcov-report/natural/stemmers/stemmer_it.js.html +0 -427
- package/coverage/lcov-report/natural/stemmers/stemmer_ja.js.html +0 -736
- package/coverage/lcov-report/natural/stemmers/stemmer_no.js.html +0 -505
- package/coverage/lcov-report/natural/stemmers/stemmer_pl.js.html +0 -496
- package/coverage/lcov-report/natural/stemmers/stemmer_ru.js.html +0 -496
- package/coverage/lcov-report/natural/tfidf/index.html +0 -350
- package/coverage/lcov-report/natural/tfidf/tfidf.js.html +0 -904
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
- package/coverage/lcov-report/natural/tokenizers/index.html +0 -519
- package/coverage/lcov-report/natural/tokenizers/regexp_tokenizer.js.html +0 -580
- package/coverage/lcov-report/natural/tokenizers/tokenizer.js.html +0 -475
- package/coverage/lcov-report/natural/tokenizers/tokenizer_ja.js.html +0 -1153
- package/coverage/lcov-report/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
- package/coverage/lcov-report/natural/transliterators/ja/index.html +0 -350
- package/coverage/lcov-report/natural/transliterators/ja/index.js.html +0 -2068
- package/coverage/lcov-report/natural/trie/index.html +0 -350
- package/coverage/lcov-report/natural/trie/trie.js.html +0 -1012
- package/coverage/lcov-report/natural/util/bag.js.html +0 -472
- package/coverage/lcov-report/natural/util/edge_weighted_digraph.js.html +0 -658
- package/coverage/lcov-report/natural/util/index.html +0 -532
- package/coverage/lcov-report/natural/util/longest_path_tree.js.html +0 -619
- package/coverage/lcov-report/natural/util/shortest_path_tree.js.html +0 -622
- package/coverage/lcov-report/natural/util/stopwords.js.html +0 -445
- package/coverage/lcov-report/natural/util/stopwords_es.js.html +0 -430
- package/coverage/lcov-report/natural/util/stopwords_fa.js.html +0 -436
- package/coverage/lcov-report/natural/util/stopwords_fr.js.html +0 -904
- package/coverage/lcov-report/natural/util/stopwords_it.js.html +0 -478
- package/coverage/lcov-report/natural/util/stopwords_ja.js.html +0 -499
- package/coverage/lcov-report/natural/util/stopwords_no.js.html +0 -445
- package/coverage/lcov-report/natural/util/stopwords_pl.js.html +0 -508
- package/coverage/lcov-report/natural/util/stopwords_ru.js.html +0 -445
- package/coverage/lcov-report/natural/util/topological.js.html +0 -616
- package/coverage/lcov-report/natural/util/utils.js.html +0 -679
- package/coverage/lcov-report/natural/wordnet/data_file.js.html +0 -580
- package/coverage/lcov-report/natural/wordnet/index.html +0 -389
- package/coverage/lcov-report/natural/wordnet/index_file.js.html +0 -742
- package/coverage/lcov-report/natural/wordnet/wordnet.js.html +0 -829
- package/coverage/lcov-report/natural/wordnet/wordnet_file.js.html +0 -529
- package/coverage/lcov-report/normalizers/index.html +0 -350
- package/coverage/lcov-report/normalizers/normalizer.js.html +0 -607
- package/coverage/lcov-report/prettify.css +0 -1
- package/coverage/lcov-report/prettify.js +0 -1
- package/coverage/lcov-report/spec/aggressive_tokenizer_es_spec.js.html +0 -430
- package/coverage/lcov-report/spec/aggressive_tokenizer_fr_spec.js.html +0 -625
- package/coverage/lcov-report/spec/aggressive_tokenizer_nl_spec.js.html +0 -430
- package/coverage/lcov-report/spec/aggressive_tokenizer_pt_spec.js.html +0 -490
- package/coverage/lcov-report/spec/aggressive_tokenizer_spec.js.html +0 -490
- package/coverage/lcov-report/spec/bayes_classifier_spec.js.html +0 -841
- package/coverage/lcov-report/spec/classifier_spec.js.html +0 -847
- package/coverage/lcov-report/spec/count_inflector_fr_spec.js.html +0 -454
- package/coverage/lcov-report/spec/count_inflector_spec.js.html +0 -538
- package/coverage/lcov-report/spec/dice_coefficient_spec.js.html +0 -400
- package/coverage/lcov-report/spec/double_metaphone_spec.js.html +0 -2548
- package/coverage/lcov-report/spec/index.html +0 -883
- package/coverage/lcov-report/spec/jaro-winkler_spec.js.html +0 -481
- package/coverage/lcov-report/spec/lancaster_stemmer_spec.js.html +0 -541
- package/coverage/lcov-report/spec/levenshtein_spec.js.html +0 -520
- package/coverage/lcov-report/spec/logistic_regression_classifier_spec.js.html +0 -688
- package/coverage/lcov-report/spec/longest_path_tree_spec.js.html +0 -628
- package/coverage/lcov-report/spec/metaphone_spec.js.html +0 -1063
- package/coverage/lcov-report/spec/ngram_spec.js.html +0 -712
- package/coverage/lcov-report/spec/ngram_zh_spec.js.html +0 -688
- package/coverage/lcov-report/spec/normalizer_ja_spec.js.html +0 -802
- package/coverage/lcov-report/spec/normalizer_no_spec.js.html +0 -655
- package/coverage/lcov-report/spec/normalizer_spec.js.html +0 -742
- package/coverage/lcov-report/spec/noun_inflector_fr_spec.js.html +0 -1750
- package/coverage/lcov-report/spec/noun_inflector_ja_spec.js.html +0 -658
- package/coverage/lcov-report/spec/noun_inflector_spec.js.html +0 -1132
- package/coverage/lcov-report/spec/porter_stemmer_fr_spec.js.html +0 -625
- package/coverage/lcov-report/spec/porter_stemmer_it_spec.js.html +0 -454
- package/coverage/lcov-report/spec/porter_stemmer_no_spec.js.html +0 -751
- package/coverage/lcov-report/spec/porter_stemmer_ru_spec.js.html +0 -511
- package/coverage/lcov-report/spec/porter_stemmer_spec.js.html +0 -733
- package/coverage/lcov-report/spec/present_verb_inflector_spec.js.html +0 -697
- package/coverage/lcov-report/spec/remove_diacritics_spec.js.html +0 -415
- package/coverage/lcov-report/spec/sentence_analyzer_spec.js.html +0 -1153
- package/coverage/lcov-report/spec/shortest_path_tree_spec.js.html +0 -718
- package/coverage/lcov-report/spec/soundex_spec.js.html +0 -784
- package/coverage/lcov-report/spec/spellcheck_spec.js.html +0 -598
- package/coverage/lcov-report/spec/stemmer_ja_spec.js.html +0 -520
- package/coverage/lcov-report/spec/tfidf_spec.js.html +0 -1288
- package/coverage/lcov-report/spec/tokenizer_ja_spec.js.html +0 -496
- package/coverage/lcov-report/spec/transliterator_ja_spec.js.html +0 -631
- package/coverage/lcov-report/spec/treebank_word_tokenizer_spec.js.html +0 -430
- package/coverage/lcov-report/spec/trie_spec.js.html +0 -940
- package/coverage/lcov-report/util/index.html +0 -350
- package/coverage/lcov-report/util/utils.js.html +0 -679
- package/coverage/lcov.info +0 -11094
- package/lib/natural/parsers/Agenda.js +0 -68
- package/lib/natural/parsers/CYKParser.js +0 -87
- package/lib/natural/parsers/CYK_Item.js +0 -159
- package/lib/natural/parsers/Chart.js +0 -211
- package/lib/natural/parsers/ChartParser.js +0 -60
- package/lib/natural/parsers/DoubleDottedItem.js +0 -226
- package/lib/natural/parsers/EarleyItem.js +0 -155
- package/lib/natural/parsers/EarleyParser.js +0 -58
- package/lib/natural/parsers/FunctionWordTagger.js +0 -84
- package/lib/natural/parsers/GoalItem.js +0 -206
- package/lib/natural/parsers/GrammarParser.js +0 -1289
- package/lib/natural/parsers/HeadCornerParser.js +0 -101
- package/lib/natural/parsers/LeftCornerParser.js +0 -58
- package/lib/natural/parsers/PEG-grammar-for-unification-grammar.txt +0 -345
|
@@ -1,904 +0,0 @@
|
|
|
1
|
-
<!doctype html>
|
|
2
|
-
<html lang="en">
|
|
3
|
-
<head>
|
|
4
|
-
<title>Code coverage report for natural/tfidf/tfidf.js</title>
|
|
5
|
-
<meta charset="utf-8">
|
|
6
|
-
|
|
7
|
-
<link rel="stylesheet" href="../../prettify.css">
|
|
8
|
-
|
|
9
|
-
<style>
|
|
10
|
-
body, html {
|
|
11
|
-
margin:0; padding: 0;
|
|
12
|
-
}
|
|
13
|
-
body {
|
|
14
|
-
font-family: Helvetica Neue, Helvetica,Arial;
|
|
15
|
-
font-size: 10pt;
|
|
16
|
-
}
|
|
17
|
-
div.header, div.footer {
|
|
18
|
-
background: #eee;
|
|
19
|
-
padding: 1em;
|
|
20
|
-
}
|
|
21
|
-
div.header {
|
|
22
|
-
z-index: 100;
|
|
23
|
-
position: fixed;
|
|
24
|
-
top: 0;
|
|
25
|
-
border-bottom: 1px solid #666;
|
|
26
|
-
width: 100%;
|
|
27
|
-
}
|
|
28
|
-
div.footer {
|
|
29
|
-
border-top: 1px solid #666;
|
|
30
|
-
}
|
|
31
|
-
div.body {
|
|
32
|
-
margin-top: 10em;
|
|
33
|
-
}
|
|
34
|
-
div.meta {
|
|
35
|
-
font-size: 90%;
|
|
36
|
-
text-align: center;
|
|
37
|
-
}
|
|
38
|
-
h1, h2, h3 {
|
|
39
|
-
font-weight: normal;
|
|
40
|
-
}
|
|
41
|
-
h1 {
|
|
42
|
-
font-size: 12pt;
|
|
43
|
-
}
|
|
44
|
-
h2 {
|
|
45
|
-
font-size: 10pt;
|
|
46
|
-
}
|
|
47
|
-
pre {
|
|
48
|
-
font-family: Consolas, Menlo, Monaco, monospace;
|
|
49
|
-
margin: 0;
|
|
50
|
-
padding: 0;
|
|
51
|
-
line-height: 14px;
|
|
52
|
-
font-size: 14px;
|
|
53
|
-
-moz-tab-size: 2;
|
|
54
|
-
-o-tab-size: 2;
|
|
55
|
-
tab-size: 2;
|
|
56
|
-
}
|
|
57
|
-
|
|
58
|
-
div.path { font-size: 110%; }
|
|
59
|
-
div.path a:link, div.path a:visited { color: #000; }
|
|
60
|
-
table.coverage { border-collapse: collapse; margin:0; padding: 0 }
|
|
61
|
-
|
|
62
|
-
table.coverage td {
|
|
63
|
-
margin: 0;
|
|
64
|
-
padding: 0;
|
|
65
|
-
color: #111;
|
|
66
|
-
vertical-align: top;
|
|
67
|
-
}
|
|
68
|
-
table.coverage td.line-count {
|
|
69
|
-
width: 50px;
|
|
70
|
-
text-align: right;
|
|
71
|
-
padding-right: 5px;
|
|
72
|
-
}
|
|
73
|
-
table.coverage td.line-coverage {
|
|
74
|
-
color: #777 !important;
|
|
75
|
-
text-align: right;
|
|
76
|
-
border-left: 1px solid #666;
|
|
77
|
-
border-right: 1px solid #666;
|
|
78
|
-
}
|
|
79
|
-
|
|
80
|
-
table.coverage td.text {
|
|
81
|
-
}
|
|
82
|
-
|
|
83
|
-
table.coverage td span.cline-any {
|
|
84
|
-
display: inline-block;
|
|
85
|
-
padding: 0 5px;
|
|
86
|
-
width: 40px;
|
|
87
|
-
}
|
|
88
|
-
table.coverage td span.cline-neutral {
|
|
89
|
-
background: #eee;
|
|
90
|
-
}
|
|
91
|
-
table.coverage td span.cline-yes {
|
|
92
|
-
background: #b5d592;
|
|
93
|
-
color: #999;
|
|
94
|
-
}
|
|
95
|
-
table.coverage td span.cline-no {
|
|
96
|
-
background: #fc8c84;
|
|
97
|
-
}
|
|
98
|
-
|
|
99
|
-
.cstat-yes { color: #111; }
|
|
100
|
-
.cstat-no { background: #fc8c84; color: #111; }
|
|
101
|
-
.fstat-no { background: #ffc520; color: #111 !important; }
|
|
102
|
-
.cbranch-no { background: yellow !important; color: #111; }
|
|
103
|
-
|
|
104
|
-
.cstat-skip { background: #ddd; color: #111; }
|
|
105
|
-
.fstat-skip { background: #ddd; color: #111 !important; }
|
|
106
|
-
.cbranch-skip { background: #ddd !important; color: #111; }
|
|
107
|
-
|
|
108
|
-
.missing-if-branch {
|
|
109
|
-
display: inline-block;
|
|
110
|
-
margin-right: 10px;
|
|
111
|
-
position: relative;
|
|
112
|
-
padding: 0 4px;
|
|
113
|
-
background: black;
|
|
114
|
-
color: yellow;
|
|
115
|
-
}
|
|
116
|
-
|
|
117
|
-
.skip-if-branch {
|
|
118
|
-
display: none;
|
|
119
|
-
margin-right: 10px;
|
|
120
|
-
position: relative;
|
|
121
|
-
padding: 0 4px;
|
|
122
|
-
background: #ccc;
|
|
123
|
-
color: white;
|
|
124
|
-
}
|
|
125
|
-
|
|
126
|
-
.missing-if-branch .typ, .skip-if-branch .typ {
|
|
127
|
-
color: inherit !important;
|
|
128
|
-
}
|
|
129
|
-
|
|
130
|
-
.entity, .metric { font-weight: bold; }
|
|
131
|
-
.metric { display: inline-block; border: 1px solid #333; padding: 0.3em; background: white; }
|
|
132
|
-
.metric small { font-size: 80%; font-weight: normal; color: #666; }
|
|
133
|
-
|
|
134
|
-
div.coverage-summary table { border-collapse: collapse; margin: 3em; font-size: 110%; }
|
|
135
|
-
div.coverage-summary td, div.coverage-summary table th { margin: 0; padding: 0.25em 1em; border-top: 1px solid #666; border-bottom: 1px solid #666; }
|
|
136
|
-
div.coverage-summary th { text-align: left; border: 1px solid #666; background: #eee; font-weight: normal; }
|
|
137
|
-
div.coverage-summary th.file { border-right: none !important; }
|
|
138
|
-
div.coverage-summary th.pic { border-left: none !important; text-align: right; }
|
|
139
|
-
div.coverage-summary th.pct { border-right: none !important; }
|
|
140
|
-
div.coverage-summary th.abs { border-left: none !important; text-align: right; }
|
|
141
|
-
div.coverage-summary td.pct { text-align: right; border-left: 1px solid #666; }
|
|
142
|
-
div.coverage-summary td.abs { text-align: right; font-size: 90%; color: #444; border-right: 1px solid #666; }
|
|
143
|
-
div.coverage-summary td.file { text-align: right; border-left: 1px solid #666; white-space: nowrap; }
|
|
144
|
-
div.coverage-summary td.pic { min-width: 120px !important; }
|
|
145
|
-
div.coverage-summary a:link { text-decoration: none; color: #000; }
|
|
146
|
-
div.coverage-summary a:visited { text-decoration: none; color: #333; }
|
|
147
|
-
div.coverage-summary a:hover { text-decoration: underline; }
|
|
148
|
-
div.coverage-summary tfoot td { border-top: 1px solid #666; }
|
|
149
|
-
|
|
150
|
-
div.coverage-summary .yui3-datatable-sort-indicator, div.coverage-summary .dummy-sort-indicator {
|
|
151
|
-
height: 10px;
|
|
152
|
-
width: 7px;
|
|
153
|
-
display: inline-block;
|
|
154
|
-
margin-left: 0.5em;
|
|
155
|
-
}
|
|
156
|
-
div.coverage-summary .yui3-datatable-sort-indicator {
|
|
157
|
-
background: url("https://yui-s.yahooapis.com/3.6.0/build/datatable-sort/assets/skins/sam/sort-arrow-sprite.png") no-repeat scroll 0 0 transparent;
|
|
158
|
-
}
|
|
159
|
-
div.coverage-summary .yui3-datatable-sorted .yui3-datatable-sort-indicator {
|
|
160
|
-
background-position: 0 -20px;
|
|
161
|
-
}
|
|
162
|
-
div.coverage-summary .yui3-datatable-sorted-desc .yui3-datatable-sort-indicator {
|
|
163
|
-
background-position: 0 -10px;
|
|
164
|
-
}
|
|
165
|
-
|
|
166
|
-
.high { background: #b5d592 !important; }
|
|
167
|
-
.medium { background: #ffe87c !important; }
|
|
168
|
-
.low { background: #fc8c84 !important; }
|
|
169
|
-
|
|
170
|
-
span.cover-fill, span.cover-empty {
|
|
171
|
-
display:inline-block;
|
|
172
|
-
border:1px solid #444;
|
|
173
|
-
background: white;
|
|
174
|
-
height: 12px;
|
|
175
|
-
}
|
|
176
|
-
span.cover-fill {
|
|
177
|
-
background: #ccc;
|
|
178
|
-
border-right: 1px solid #444;
|
|
179
|
-
}
|
|
180
|
-
span.cover-empty {
|
|
181
|
-
background: white;
|
|
182
|
-
border-left: none;
|
|
183
|
-
}
|
|
184
|
-
span.cover-full {
|
|
185
|
-
border-right: none !important;
|
|
186
|
-
}
|
|
187
|
-
pre.prettyprint {
|
|
188
|
-
border: none !important;
|
|
189
|
-
padding: 0 !important;
|
|
190
|
-
margin: 0 !important;
|
|
191
|
-
}
|
|
192
|
-
.com { color: #999 !important; }
|
|
193
|
-
.ignore-none { color: #999; font-weight: normal; }
|
|
194
|
-
|
|
195
|
-
</style>
|
|
196
|
-
</head>
|
|
197
|
-
<body>
|
|
198
|
-
<div class="header low">
|
|
199
|
-
<h1>Code coverage report for <span class="entity">natural/tfidf/tfidf.js</span></h1>
|
|
200
|
-
<h2>
|
|
201
|
-
|
|
202
|
-
Statements: <span class="metric">18.07% <small>(15 / 83)</small></span>
|
|
203
|
-
|
|
204
|
-
|
|
205
|
-
Branches: <span class="metric">0% <small>(0 / 56)</small></span>
|
|
206
|
-
|
|
207
|
-
|
|
208
|
-
Functions: <span class="metric">0% <small>(0 / 16)</small></span>
|
|
209
|
-
|
|
210
|
-
|
|
211
|
-
Lines: <span class="metric">18.52% <small>(15 / 81)</small></span>
|
|
212
|
-
|
|
213
|
-
Ignored: <span class="metric"><span class="ignore-none">none</span></span>
|
|
214
|
-
</h2>
|
|
215
|
-
<div class="path"><a href="../../index.html">All files</a> » <a href="index.html">natural/tfidf/</a> » tfidf.js</div>
|
|
216
|
-
</div>
|
|
217
|
-
<div class="body">
|
|
218
|
-
<pre><table class="coverage">
|
|
219
|
-
<tr><td class="line-count">1
|
|
220
|
-
2
|
|
221
|
-
3
|
|
222
|
-
4
|
|
223
|
-
5
|
|
224
|
-
6
|
|
225
|
-
7
|
|
226
|
-
8
|
|
227
|
-
9
|
|
228
|
-
10
|
|
229
|
-
11
|
|
230
|
-
12
|
|
231
|
-
13
|
|
232
|
-
14
|
|
233
|
-
15
|
|
234
|
-
16
|
|
235
|
-
17
|
|
236
|
-
18
|
|
237
|
-
19
|
|
238
|
-
20
|
|
239
|
-
21
|
|
240
|
-
22
|
|
241
|
-
23
|
|
242
|
-
24
|
|
243
|
-
25
|
|
244
|
-
26
|
|
245
|
-
27
|
|
246
|
-
28
|
|
247
|
-
29
|
|
248
|
-
30
|
|
249
|
-
31
|
|
250
|
-
32
|
|
251
|
-
33
|
|
252
|
-
34
|
|
253
|
-
35
|
|
254
|
-
36
|
|
255
|
-
37
|
|
256
|
-
38
|
|
257
|
-
39
|
|
258
|
-
40
|
|
259
|
-
41
|
|
260
|
-
42
|
|
261
|
-
43
|
|
262
|
-
44
|
|
263
|
-
45
|
|
264
|
-
46
|
|
265
|
-
47
|
|
266
|
-
48
|
|
267
|
-
49
|
|
268
|
-
50
|
|
269
|
-
51
|
|
270
|
-
52
|
|
271
|
-
53
|
|
272
|
-
54
|
|
273
|
-
55
|
|
274
|
-
56
|
|
275
|
-
57
|
|
276
|
-
58
|
|
277
|
-
59
|
|
278
|
-
60
|
|
279
|
-
61
|
|
280
|
-
62
|
|
281
|
-
63
|
|
282
|
-
64
|
|
283
|
-
65
|
|
284
|
-
66
|
|
285
|
-
67
|
|
286
|
-
68
|
|
287
|
-
69
|
|
288
|
-
70
|
|
289
|
-
71
|
|
290
|
-
72
|
|
291
|
-
73
|
|
292
|
-
74
|
|
293
|
-
75
|
|
294
|
-
76
|
|
295
|
-
77
|
|
296
|
-
78
|
|
297
|
-
79
|
|
298
|
-
80
|
|
299
|
-
81
|
|
300
|
-
82
|
|
301
|
-
83
|
|
302
|
-
84
|
|
303
|
-
85
|
|
304
|
-
86
|
|
305
|
-
87
|
|
306
|
-
88
|
|
307
|
-
89
|
|
308
|
-
90
|
|
309
|
-
91
|
|
310
|
-
92
|
|
311
|
-
93
|
|
312
|
-
94
|
|
313
|
-
95
|
|
314
|
-
96
|
|
315
|
-
97
|
|
316
|
-
98
|
|
317
|
-
99
|
|
318
|
-
100
|
|
319
|
-
101
|
|
320
|
-
102
|
|
321
|
-
103
|
|
322
|
-
104
|
|
323
|
-
105
|
|
324
|
-
106
|
|
325
|
-
107
|
|
326
|
-
108
|
|
327
|
-
109
|
|
328
|
-
110
|
|
329
|
-
111
|
|
330
|
-
112
|
|
331
|
-
113
|
|
332
|
-
114
|
|
333
|
-
115
|
|
334
|
-
116
|
|
335
|
-
117
|
|
336
|
-
118
|
|
337
|
-
119
|
|
338
|
-
120
|
|
339
|
-
121
|
|
340
|
-
122
|
|
341
|
-
123
|
|
342
|
-
124
|
|
343
|
-
125
|
|
344
|
-
126
|
|
345
|
-
127
|
|
346
|
-
128
|
|
347
|
-
129
|
|
348
|
-
130
|
|
349
|
-
131
|
|
350
|
-
132
|
|
351
|
-
133
|
|
352
|
-
134
|
|
353
|
-
135
|
|
354
|
-
136
|
|
355
|
-
137
|
|
356
|
-
138
|
|
357
|
-
139
|
|
358
|
-
140
|
|
359
|
-
141
|
|
360
|
-
142
|
|
361
|
-
143
|
|
362
|
-
144
|
|
363
|
-
145
|
|
364
|
-
146
|
|
365
|
-
147
|
|
366
|
-
148
|
|
367
|
-
149
|
|
368
|
-
150
|
|
369
|
-
151
|
|
370
|
-
152
|
|
371
|
-
153
|
|
372
|
-
154
|
|
373
|
-
155
|
|
374
|
-
156
|
|
375
|
-
157
|
|
376
|
-
158
|
|
377
|
-
159
|
|
378
|
-
160
|
|
379
|
-
161
|
|
380
|
-
162
|
|
381
|
-
163
|
|
382
|
-
164
|
|
383
|
-
165
|
|
384
|
-
166
|
|
385
|
-
167
|
|
386
|
-
168
|
|
387
|
-
169
|
|
388
|
-
170
|
|
389
|
-
171
|
|
390
|
-
172
|
|
391
|
-
173
|
|
392
|
-
174
|
|
393
|
-
175
|
|
394
|
-
176
|
|
395
|
-
177
|
|
396
|
-
178
|
|
397
|
-
179
|
|
398
|
-
180
|
|
399
|
-
181
|
|
400
|
-
182
|
|
401
|
-
183
|
|
402
|
-
184
|
|
403
|
-
185
|
|
404
|
-
186
|
|
405
|
-
187
|
|
406
|
-
188
|
|
407
|
-
189
|
|
408
|
-
190
|
|
409
|
-
191
|
|
410
|
-
192
|
|
411
|
-
193
|
|
412
|
-
194
|
|
413
|
-
195</td><td class="line-coverage"><span class="cline-any cline-neutral"> </span>
|
|
414
|
-
<span class="cline-any cline-neutral"> </span>
|
|
415
|
-
<span class="cline-any cline-neutral"> </span>
|
|
416
|
-
<span class="cline-any cline-neutral"> </span>
|
|
417
|
-
<span class="cline-any cline-neutral"> </span>
|
|
418
|
-
<span class="cline-any cline-neutral"> </span>
|
|
419
|
-
<span class="cline-any cline-neutral"> </span>
|
|
420
|
-
<span class="cline-any cline-neutral"> </span>
|
|
421
|
-
<span class="cline-any cline-neutral"> </span>
|
|
422
|
-
<span class="cline-any cline-neutral"> </span>
|
|
423
|
-
<span class="cline-any cline-neutral"> </span>
|
|
424
|
-
<span class="cline-any cline-neutral"> </span>
|
|
425
|
-
<span class="cline-any cline-neutral"> </span>
|
|
426
|
-
<span class="cline-any cline-neutral"> </span>
|
|
427
|
-
<span class="cline-any cline-neutral"> </span>
|
|
428
|
-
<span class="cline-any cline-neutral"> </span>
|
|
429
|
-
<span class="cline-any cline-neutral"> </span>
|
|
430
|
-
<span class="cline-any cline-neutral"> </span>
|
|
431
|
-
<span class="cline-any cline-neutral"> </span>
|
|
432
|
-
<span class="cline-any cline-neutral"> </span>
|
|
433
|
-
<span class="cline-any cline-neutral"> </span>
|
|
434
|
-
<span class="cline-any cline-neutral"> </span>
|
|
435
|
-
<span class="cline-any cline-yes">1</span>
|
|
436
|
-
<span class="cline-any cline-neutral"> </span>
|
|
437
|
-
<span class="cline-any cline-neutral"> </span>
|
|
438
|
-
<span class="cline-any cline-neutral"> </span>
|
|
439
|
-
<span class="cline-any cline-neutral"> </span>
|
|
440
|
-
<span class="cline-any cline-neutral"> </span>
|
|
441
|
-
<span class="cline-any cline-yes">1</span>
|
|
442
|
-
<span class="cline-any cline-no"> </span>
|
|
443
|
-
<span class="cline-any cline-neutral"> </span>
|
|
444
|
-
<span class="cline-any cline-no"> </span>
|
|
445
|
-
<span class="cline-any cline-no"> </span>
|
|
446
|
-
<span class="cline-any cline-no"> </span>
|
|
447
|
-
<span class="cline-any cline-no"> </span>
|
|
448
|
-
<span class="cline-any cline-no"> </span>
|
|
449
|
-
<span class="cline-any cline-no"> </span>
|
|
450
|
-
<span class="cline-any cline-neutral"> </span>
|
|
451
|
-
<span class="cline-any cline-neutral"> </span>
|
|
452
|
-
<span class="cline-any cline-no"> </span>
|
|
453
|
-
<span class="cline-any cline-neutral"> </span>
|
|
454
|
-
<span class="cline-any cline-no"> </span>
|
|
455
|
-
<span class="cline-any cline-no"> </span>
|
|
456
|
-
<span class="cline-any cline-no"> </span>
|
|
457
|
-
<span class="cline-any cline-no"> </span>
|
|
458
|
-
<span class="cline-any cline-neutral"> </span>
|
|
459
|
-
<span class="cline-any cline-neutral"> </span>
|
|
460
|
-
<span class="cline-any cline-neutral"> </span>
|
|
461
|
-
<span class="cline-any cline-yes">1</span>
|
|
462
|
-
<span class="cline-any cline-no"> </span>
|
|
463
|
-
<span class="cline-any cline-neutral"> </span>
|
|
464
|
-
<span class="cline-any cline-neutral"> </span>
|
|
465
|
-
<span class="cline-any cline-yes">1</span>
|
|
466
|
-
<span class="cline-any cline-no"> </span>
|
|
467
|
-
<span class="cline-any cline-neutral"> </span>
|
|
468
|
-
<span class="cline-any cline-neutral"> </span>
|
|
469
|
-
<span class="cline-any cline-yes">1</span>
|
|
470
|
-
<span class="cline-any cline-no"> </span>
|
|
471
|
-
<span class="cline-any cline-no"> </span>
|
|
472
|
-
<span class="cline-any cline-neutral"> </span>
|
|
473
|
-
<span class="cline-any cline-no"> </span>
|
|
474
|
-
<span class="cline-any cline-neutral"> </span>
|
|
475
|
-
<span class="cline-any cline-no"> </span>
|
|
476
|
-
<span class="cline-any cline-neutral"> </span>
|
|
477
|
-
<span class="cline-any cline-neutral"> </span>
|
|
478
|
-
<span class="cline-any cline-neutral"> </span>
|
|
479
|
-
<span class="cline-any cline-yes">1</span>
|
|
480
|
-
<span class="cline-any cline-no"> </span>
|
|
481
|
-
<span class="cline-any cline-no"> </span>
|
|
482
|
-
<span class="cline-any cline-no"> </span>
|
|
483
|
-
<span class="cline-any cline-neutral"> </span>
|
|
484
|
-
<span class="cline-any cline-neutral"> </span>
|
|
485
|
-
<span class="cline-any cline-neutral"> </span>
|
|
486
|
-
<span class="cline-any cline-neutral"> </span>
|
|
487
|
-
<span class="cline-any cline-neutral"> </span>
|
|
488
|
-
<span class="cline-any cline-neutral"> </span>
|
|
489
|
-
<span class="cline-any cline-neutral"> </span>
|
|
490
|
-
<span class="cline-any cline-neutral"> </span>
|
|
491
|
-
<span class="cline-any cline-neutral"> </span>
|
|
492
|
-
<span class="cline-any cline-neutral"> </span>
|
|
493
|
-
<span class="cline-any cline-neutral"> </span>
|
|
494
|
-
<span class="cline-any cline-no"> </span>
|
|
495
|
-
<span class="cline-any cline-neutral"> </span>
|
|
496
|
-
<span class="cline-any cline-no"> </span>
|
|
497
|
-
<span class="cline-any cline-neutral"> </span>
|
|
498
|
-
<span class="cline-any cline-neutral"> </span>
|
|
499
|
-
<span class="cline-any cline-yes">1</span>
|
|
500
|
-
<span class="cline-any cline-yes">1</span>
|
|
501
|
-
<span class="cline-any cline-neutral"> </span>
|
|
502
|
-
<span class="cline-any cline-yes">1</span>
|
|
503
|
-
<span class="cline-any cline-neutral"> </span>
|
|
504
|
-
<span class="cline-any cline-neutral"> </span>
|
|
505
|
-
<span class="cline-any cline-neutral"> </span>
|
|
506
|
-
<span class="cline-any cline-no"> </span>
|
|
507
|
-
<span class="cline-any cline-no"> </span>
|
|
508
|
-
<span class="cline-any cline-neutral"> </span>
|
|
509
|
-
<span class="cline-any cline-no"> </span>
|
|
510
|
-
<span class="cline-any cline-no"> </span>
|
|
511
|
-
<span class="cline-any cline-neutral"> </span>
|
|
512
|
-
<span class="cline-any cline-neutral"> </span>
|
|
513
|
-
<span class="cline-any cline-no"> </span>
|
|
514
|
-
<span class="cline-any cline-neutral"> </span>
|
|
515
|
-
<span class="cline-any cline-neutral"> </span>
|
|
516
|
-
<span class="cline-any cline-no"> </span>
|
|
517
|
-
<span class="cline-any cline-no"> </span>
|
|
518
|
-
<span class="cline-any cline-neutral"> </span>
|
|
519
|
-
<span class="cline-any cline-neutral"> </span>
|
|
520
|
-
<span class="cline-any cline-neutral"> </span>
|
|
521
|
-
<span class="cline-any cline-neutral"> </span>
|
|
522
|
-
<span class="cline-any cline-yes">1</span>
|
|
523
|
-
<span class="cline-any cline-no"> </span>
|
|
524
|
-
<span class="cline-any cline-neutral"> </span>
|
|
525
|
-
<span class="cline-any cline-neutral"> </span>
|
|
526
|
-
<span class="cline-any cline-no"> </span>
|
|
527
|
-
<span class="cline-any cline-no"> </span>
|
|
528
|
-
<span class="cline-any cline-neutral"> </span>
|
|
529
|
-
<span class="cline-any cline-neutral"> </span>
|
|
530
|
-
<span class="cline-any cline-neutral"> </span>
|
|
531
|
-
<span class="cline-any cline-no"> </span>
|
|
532
|
-
<span class="cline-any cline-neutral"> </span>
|
|
533
|
-
<span class="cline-any cline-neutral"> </span>
|
|
534
|
-
<span class="cline-any cline-neutral"> </span>
|
|
535
|
-
<span class="cline-any cline-no"> </span>
|
|
536
|
-
<span class="cline-any cline-neutral"> </span>
|
|
537
|
-
<span class="cline-any cline-neutral"> </span>
|
|
538
|
-
<span class="cline-any cline-neutral"> </span>
|
|
539
|
-
<span class="cline-any cline-neutral"> </span>
|
|
540
|
-
<span class="cline-any cline-neutral"> </span>
|
|
541
|
-
<span class="cline-any cline-yes">1</span>
|
|
542
|
-
<span class="cline-any cline-no"> </span>
|
|
543
|
-
<span class="cline-any cline-no"> </span>
|
|
544
|
-
<span class="cline-any cline-no"> </span>
|
|
545
|
-
<span class="cline-any cline-no"> </span>
|
|
546
|
-
<span class="cline-any cline-neutral"> </span>
|
|
547
|
-
<span class="cline-any cline-no"> </span>
|
|
548
|
-
<span class="cline-any cline-no"> </span>
|
|
549
|
-
<span class="cline-any cline-neutral"> </span>
|
|
550
|
-
<span class="cline-any cline-neutral"> </span>
|
|
551
|
-
<span class="cline-any cline-no"> </span>
|
|
552
|
-
<span class="cline-any cline-no"> </span>
|
|
553
|
-
<span class="cline-any cline-neutral"> </span>
|
|
554
|
-
<span class="cline-any cline-neutral"> </span>
|
|
555
|
-
<span class="cline-any cline-neutral"> </span>
|
|
556
|
-
<span class="cline-any cline-no"> </span>
|
|
557
|
-
<span class="cline-any cline-neutral"> </span>
|
|
558
|
-
<span class="cline-any cline-neutral"> </span>
|
|
559
|
-
<span class="cline-any cline-neutral"> </span>
|
|
560
|
-
<span class="cline-any cline-no"> </span>
|
|
561
|
-
<span class="cline-any cline-neutral"> </span>
|
|
562
|
-
<span class="cline-any cline-neutral"> </span>
|
|
563
|
-
<span class="cline-any cline-neutral"> </span>
|
|
564
|
-
<span class="cline-any cline-yes">1</span>
|
|
565
|
-
<span class="cline-any cline-no"> </span>
|
|
566
|
-
<span class="cline-any cline-neutral"> </span>
|
|
567
|
-
<span class="cline-any cline-no"> </span>
|
|
568
|
-
<span class="cline-any cline-no"> </span>
|
|
569
|
-
<span class="cline-any cline-neutral"> </span>
|
|
570
|
-
<span class="cline-any cline-no"> </span>
|
|
571
|
-
<span class="cline-any cline-no"> </span>
|
|
572
|
-
<span class="cline-any cline-no"> </span>
|
|
573
|
-
<span class="cline-any cline-no"> </span>
|
|
574
|
-
<span class="cline-any cline-neutral"> </span>
|
|
575
|
-
<span class="cline-any cline-neutral"> </span>
|
|
576
|
-
<span class="cline-any cline-neutral"> </span>
|
|
577
|
-
<span class="cline-any cline-yes">1</span>
|
|
578
|
-
<span class="cline-any cline-no"> </span>
|
|
579
|
-
<span class="cline-any cline-neutral"> </span>
|
|
580
|
-
<span class="cline-any cline-no"> </span>
|
|
581
|
-
<span class="cline-any cline-no"> </span>
|
|
582
|
-
<span class="cline-any cline-no"> </span>
|
|
583
|
-
<span class="cline-any cline-neutral"> </span>
|
|
584
|
-
<span class="cline-any cline-neutral"> </span>
|
|
585
|
-
<span class="cline-any cline-no"> </span>
|
|
586
|
-
<span class="cline-any cline-neutral"> </span>
|
|
587
|
-
<span class="cline-any cline-neutral"> </span>
|
|
588
|
-
<span class="cline-any cline-yes">1</span>
|
|
589
|
-
<span class="cline-any cline-no"> </span>
|
|
590
|
-
<span class="cline-any cline-neutral"> </span>
|
|
591
|
-
<span class="cline-any cline-no"> </span>
|
|
592
|
-
<span class="cline-any cline-no"> </span>
|
|
593
|
-
<span class="cline-any cline-neutral"> </span>
|
|
594
|
-
<span class="cline-any cline-no"> </span>
|
|
595
|
-
<span class="cline-any cline-no"> </span>
|
|
596
|
-
<span class="cline-any cline-neutral"> </span>
|
|
597
|
-
<span class="cline-any cline-neutral"> </span>
|
|
598
|
-
<span class="cline-any cline-no"> </span>
|
|
599
|
-
<span class="cline-any cline-neutral"> </span>
|
|
600
|
-
<span class="cline-any cline-neutral"> </span>
|
|
601
|
-
<span class="cline-any cline-neutral"> </span>
|
|
602
|
-
<span class="cline-any cline-yes">1</span>
|
|
603
|
-
<span class="cline-any cline-no"> </span>
|
|
604
|
-
<span class="cline-any cline-no"> </span>
|
|
605
|
-
<span class="cline-any cline-no"> </span>
|
|
606
|
-
<span class="cline-any cline-neutral"> </span>
|
|
607
|
-
<span class="cline-any cline-neutral"> </span></td><td class="text"><pre class="prettyprint lang-js">/*
|
|
608
|
-
Copyright (c) 2011, Rob Ellis, Chris Umbel
|
|
609
|
-
|
|
610
|
-
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
611
|
-
of this software and associated documentation files (the "Software"), to deal
|
|
612
|
-
in the Software without restriction, including without limitation the rights
|
|
613
|
-
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
614
|
-
copies of the Software, and to permit persons to whom the Software is
|
|
615
|
-
furnished to do so, subject to the following conditions:
|
|
616
|
-
|
|
617
|
-
The above copyright notice and this permission notice shall be included in
|
|
618
|
-
all copies or substantial portions of the Software.
|
|
619
|
-
|
|
620
|
-
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
621
|
-
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
622
|
-
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
623
|
-
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
624
|
-
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
625
|
-
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
|
|
626
|
-
THE SOFTWARE.
|
|
627
|
-
*/
|
|
628
|
-
|
|
629
|
-
var _ = require("underscore")._,
|
|
630
|
-
Tokenizer = require('../tokenizers/regexp_tokenizer').WordTokenizer,
|
|
631
|
-
tokenizer = new Tokenizer(),
|
|
632
|
-
stopwords = require('../util/stopwords').words,
|
|
633
|
-
fs = require('fs');
|
|
634
|
-
|
|
635
|
-
<span class="fstat-no" title="function not covered" >function buildDocument(text, key) {</span>
|
|
636
|
-
<span class="cstat-no" title="statement not covered" > var stopOut;</span>
|
|
637
|
-
|
|
638
|
-
<span class="cstat-no" title="statement not covered" > if(typeof text === 'string') {</span>
|
|
639
|
-
<span class="cstat-no" title="statement not covered" > text = tokenizer.tokenize(text.toLowerCase());</span>
|
|
640
|
-
<span class="cstat-no" title="statement not covered" > stopOut = true;</span>
|
|
641
|
-
} else <span class="cstat-no" title="statement not covered" >if(!_.isArray(text)) {</span>
|
|
642
|
-
<span class="cstat-no" title="statement not covered" > stopOut = false;</span>
|
|
643
|
-
<span class="cstat-no" title="statement not covered" > return text;</span>
|
|
644
|
-
}
|
|
645
|
-
|
|
646
|
-
<span class="cstat-no" title="statement not covered" > return text.reduce(<span class="fstat-no" title="function not covered" >function(document, term) {</span></span>
|
|
647
|
-
// next line solves https://github.com/NaturalNode/natural/issues/119
|
|
648
|
-
<span class="cstat-no" title="statement not covered" > if(typeof document[term] === 'function') <span class="cstat-no" title="statement not covered" >document[term] = 0;</span></span>
|
|
649
|
-
<span class="cstat-no" title="statement not covered" > if(!stopOut || stopwords.indexOf(term) < 0)</span>
|
|
650
|
-
<span class="cstat-no" title="statement not covered" > document[term] = (document[term] ? document[term] + 1 : 1);</span>
|
|
651
|
-
<span class="cstat-no" title="statement not covered" > return document;</span>
|
|
652
|
-
}, {__key: key});
|
|
653
|
-
}
|
|
654
|
-
|
|
655
|
-
<span class="fstat-no" title="function not covered" >function tf(term, document) {</span>
|
|
656
|
-
<span class="cstat-no" title="statement not covered" > return document[term] ? document[term]: 0;</span>
|
|
657
|
-
}
|
|
658
|
-
|
|
659
|
-
<span class="fstat-no" title="function not covered" >function documentHasTerm(term, document) {</span>
|
|
660
|
-
<span class="cstat-no" title="statement not covered" > return document[term] && document[term] > 0;</span>
|
|
661
|
-
}
|
|
662
|
-
|
|
663
|
-
<span class="fstat-no" title="function not covered" >function TfIdf(deserialized) {</span>
|
|
664
|
-
<span class="cstat-no" title="statement not covered" > if(deserialized)</span>
|
|
665
|
-
<span class="cstat-no" title="statement not covered" > this.documents = deserialized.documents;</span>
|
|
666
|
-
else
|
|
667
|
-
<span class="cstat-no" title="statement not covered" > this.documents = [];</span>
|
|
668
|
-
|
|
669
|
-
<span class="cstat-no" title="statement not covered" > this._idfCache = {};</span>
|
|
670
|
-
}
|
|
671
|
-
|
|
672
|
-
// backwards compatibility for < node 0.10
|
|
673
|
-
<span class="fstat-no" title="function not covered" >function isEncoding(encoding) {</span>
|
|
674
|
-
<span class="cstat-no" title="statement not covered" > if (typeof Buffer.isEncoding !== 'undefined')</span>
|
|
675
|
-
<span class="cstat-no" title="statement not covered" > return Buffer.isEncoding(encoding);</span>
|
|
676
|
-
<span class="cstat-no" title="statement not covered" > switch ((encoding + '').toLowerCase()) {</span>
|
|
677
|
-
case 'hex':
|
|
678
|
-
case 'utf8':
|
|
679
|
-
case 'utf-8':
|
|
680
|
-
case 'ascii':
|
|
681
|
-
case 'binary':
|
|
682
|
-
case 'base64':
|
|
683
|
-
case 'ucs2':
|
|
684
|
-
case 'ucs-2':
|
|
685
|
-
case 'utf16le':
|
|
686
|
-
case 'utf-16le':
|
|
687
|
-
case 'raw':
|
|
688
|
-
<span class="cstat-no" title="statement not covered" > return true;</span>
|
|
689
|
-
}
|
|
690
|
-
<span class="cstat-no" title="statement not covered" > return false;</span>
|
|
691
|
-
}
|
|
692
|
-
|
|
693
|
-
module.exports = TfIdf;
|
|
694
|
-
TfIdf.tf = tf;
|
|
695
|
-
|
|
696
|
-
TfIdf.prototype.idf = <span class="fstat-no" title="function not covered" >function(term, force) {</span>
|
|
697
|
-
|
|
698
|
-
// Lookup the term in the New term-IDF caching,
|
|
699
|
-
// this will cut search times down exponentially on large document sets.
|
|
700
|
-
<span class="cstat-no" title="statement not covered" > if(this._idfCache[term] && this._idfCache.hasOwnProperty(term) && force !== true)</span>
|
|
701
|
-
<span class="cstat-no" title="statement not covered" > return this._idfCache[term];</span>
|
|
702
|
-
|
|
703
|
-
<span class="cstat-no" title="statement not covered" > var docsWithTerm = this.documents.reduce(<span class="fstat-no" title="function not covered" >function(count, document) {</span></span>
|
|
704
|
-
<span class="cstat-no" title="statement not covered" > return count + (documentHasTerm(term, document) ? 1 : 0);</span>
|
|
705
|
-
}, 0);
|
|
706
|
-
|
|
707
|
-
<span class="cstat-no" title="statement not covered" > var idf = Math.log((this.documents.length) / (docsWithTerm));</span>
|
|
708
|
-
|
|
709
|
-
// Add the idf to the term cache and return it
|
|
710
|
-
<span class="cstat-no" title="statement not covered" > this._idfCache[term] = idf;</span>
|
|
711
|
-
<span class="cstat-no" title="statement not covered" > return idf;</span>
|
|
712
|
-
};
|
|
713
|
-
|
|
714
|
-
// If restoreCache is set to true, all terms idf scores currently cached will be recomputed.
|
|
715
|
-
// Otherwise, the cache will just be wiped clean
|
|
716
|
-
TfIdf.prototype.addDocument = <span class="fstat-no" title="function not covered" >function(document, key, restoreCache) {</span>
|
|
717
|
-
<span class="cstat-no" title="statement not covered" > this.documents.push(buildDocument(document, key));</span>
|
|
718
|
-
|
|
719
|
-
// make sure the cache is invalidated when new documents arrive
|
|
720
|
-
<span class="cstat-no" title="statement not covered" > if(restoreCache === true) {</span>
|
|
721
|
-
<span class="cstat-no" title="statement not covered" > for(var term in this._idfCache) {</span>
|
|
722
|
-
// invoking idf with the force option set will
|
|
723
|
-
// force a recomputation of the idf, and it will
|
|
724
|
-
// automatically refresh the cache value.
|
|
725
|
-
<span class="cstat-no" title="statement not covered" > this.idf(term, true);</span>
|
|
726
|
-
}
|
|
727
|
-
}
|
|
728
|
-
else {
|
|
729
|
-
<span class="cstat-no" title="statement not covered" > this._idfCache = {};</span>
|
|
730
|
-
}
|
|
731
|
-
};
|
|
732
|
-
|
|
733
|
-
// If restoreCache is set to true, all terms idf scores currently cached will be recomputed.
|
|
734
|
-
// Otherwise, the cache will just be wiped clean
|
|
735
|
-
TfIdf.prototype.addFileSync = <span class="fstat-no" title="function not covered" >function(path, encoding, key, restoreCache) {</span>
|
|
736
|
-
<span class="cstat-no" title="statement not covered" > if(!encoding)</span>
|
|
737
|
-
<span class="cstat-no" title="statement not covered" > encoding = 'utf8';</span>
|
|
738
|
-
<span class="cstat-no" title="statement not covered" > if(!isEncoding(encoding))</span>
|
|
739
|
-
<span class="cstat-no" title="statement not covered" > throw new Error('Invalid encoding: ' + encoding);</span>
|
|
740
|
-
|
|
741
|
-
<span class="cstat-no" title="statement not covered" > var document = fs.readFileSync(path, encoding);</span>
|
|
742
|
-
<span class="cstat-no" title="statement not covered" > this.documents.push(buildDocument(document, key));</span>
|
|
743
|
-
|
|
744
|
-
// make sure the cache is invalidated when new documents arrive
|
|
745
|
-
<span class="cstat-no" title="statement not covered" > if(restoreCache === true) {</span>
|
|
746
|
-
<span class="cstat-no" title="statement not covered" > for(var term in this._idfCache) {</span>
|
|
747
|
-
// invoking idf with the force option set will
|
|
748
|
-
// force a recomputation of the idf, and it will
|
|
749
|
-
// automatically refresh the cache value.
|
|
750
|
-
<span class="cstat-no" title="statement not covered" > this.idf(term, true);</span>
|
|
751
|
-
}
|
|
752
|
-
}
|
|
753
|
-
else {
|
|
754
|
-
<span class="cstat-no" title="statement not covered" > this._idfCache = {};</span>
|
|
755
|
-
}
|
|
756
|
-
};
|
|
757
|
-
|
|
758
|
-
TfIdf.prototype.tfidf = <span class="fstat-no" title="function not covered" >function(terms, d) {</span>
|
|
759
|
-
<span class="cstat-no" title="statement not covered" > var _this = this;</span>
|
|
760
|
-
|
|
761
|
-
<span class="cstat-no" title="statement not covered" > if(!_.isArray(terms))</span>
|
|
762
|
-
<span class="cstat-no" title="statement not covered" > terms = tokenizer.tokenize(terms.toString().toLowerCase());</span>
|
|
763
|
-
|
|
764
|
-
<span class="cstat-no" title="statement not covered" > return terms.reduce(<span class="fstat-no" title="function not covered" >function(value, term) {</span></span>
|
|
765
|
-
<span class="cstat-no" title="statement not covered" > var idf = _this.idf(term);</span>
|
|
766
|
-
<span class="cstat-no" title="statement not covered" > idf = idf === Infinity ? 0 : idf;</span>
|
|
767
|
-
<span class="cstat-no" title="statement not covered" > return value + (tf(term, _this.documents[d]) * idf);</span>
|
|
768
|
-
}, 0.0);
|
|
769
|
-
};
|
|
770
|
-
|
|
771
|
-
TfIdf.prototype.listTerms = <span class="fstat-no" title="function not covered" >function(d) {</span>
|
|
772
|
-
<span class="cstat-no" title="statement not covered" > var terms = [];</span>
|
|
773
|
-
|
|
774
|
-
<span class="cstat-no" title="statement not covered" > for(var term in this.documents[d]) {</span>
|
|
775
|
-
<span class="cstat-no" title="statement not covered" > if(term != '__key')</span>
|
|
776
|
-
<span class="cstat-no" title="statement not covered" > terms.push({term: term, tfidf: this.tfidf(term, d)});</span>
|
|
777
|
-
}
|
|
778
|
-
|
|
779
|
-
<span class="cstat-no" title="statement not covered" > return terms.sort(<span class="fstat-no" title="function not covered" >function(x, y) {</span> <span class="cstat-no" title="statement not covered" >return y.tfidf - x.tfidf; </span>});</span>
|
|
780
|
-
};
|
|
781
|
-
|
|
782
|
-
TfIdf.prototype.tfidfs = <span class="fstat-no" title="function not covered" >function(terms, callback) {</span>
|
|
783
|
-
<span class="cstat-no" title="statement not covered" > var tfidfs = new Array(this.documents.length);</span>
|
|
784
|
-
|
|
785
|
-
<span class="cstat-no" title="statement not covered" > for(var i = 0; i < this.documents.length; i++) {</span>
|
|
786
|
-
<span class="cstat-no" title="statement not covered" > tfidfs[i] = this.tfidf(terms, i);</span>
|
|
787
|
-
|
|
788
|
-
<span class="cstat-no" title="statement not covered" > if(callback)</span>
|
|
789
|
-
<span class="cstat-no" title="statement not covered" > callback(i, tfidfs[i], this.documents[i].__key);</span>
|
|
790
|
-
}
|
|
791
|
-
|
|
792
|
-
<span class="cstat-no" title="statement not covered" > return tfidfs;</span>
|
|
793
|
-
};
|
|
794
|
-
|
|
795
|
-
// Define a tokenizer other than the default "WordTokenizer"
|
|
796
|
-
TfIdf.prototype.setTokenizer = <span class="fstat-no" title="function not covered" >function(t) {</span>
|
|
797
|
-
<span class="cstat-no" title="statement not covered" > if(!_.isFunction(t.tokenize))</span>
|
|
798
|
-
<span class="cstat-no" title="statement not covered" > throw new Error('Expected a valid Tokenizer');</span>
|
|
799
|
-
<span class="cstat-no" title="statement not covered" > tokenizer = t;</span>
|
|
800
|
-
};
|
|
801
|
-
</pre></td></tr>
|
|
802
|
-
</table></pre>
|
|
803
|
-
|
|
804
|
-
</div>
|
|
805
|
-
<div class="footer">
|
|
806
|
-
<div class="meta">Generated by <a href="http://istanbul-js.org/" target="_blank">istanbul</a> at Mon Dec 22 2014 18:01:01 GMT-0500 (EST)</div>
|
|
807
|
-
</div>
|
|
808
|
-
|
|
809
|
-
<script src="../../prettify.js"></script>
|
|
810
|
-
|
|
811
|
-
<script src="https://yui-s.yahooapis.com/3.6.0/build/yui/yui-min.js"></script>
|
|
812
|
-
<script>
|
|
813
|
-
|
|
814
|
-
YUI().use('datatable', function (Y) {
|
|
815
|
-
|
|
816
|
-
var formatters = {
|
|
817
|
-
pct: function (o) {
|
|
818
|
-
o.className += o.record.get('classes')[o.column.key];
|
|
819
|
-
try {
|
|
820
|
-
return o.value.toFixed(2) + '%';
|
|
821
|
-
} catch (ex) { return o.value + '%'; }
|
|
822
|
-
},
|
|
823
|
-
html: function (o) {
|
|
824
|
-
o.className += o.record.get('classes')[o.column.key];
|
|
825
|
-
return o.record.get(o.column.key + '_html');
|
|
826
|
-
}
|
|
827
|
-
},
|
|
828
|
-
defaultFormatter = function (o) {
|
|
829
|
-
o.className += o.record.get('classes')[o.column.key];
|
|
830
|
-
return o.value;
|
|
831
|
-
};
|
|
832
|
-
|
|
833
|
-
function getColumns(theadNode) {
|
|
834
|
-
var colNodes = theadNode.all('tr th'),
|
|
835
|
-
cols = [],
|
|
836
|
-
col;
|
|
837
|
-
colNodes.each(function (colNode) {
|
|
838
|
-
col = {
|
|
839
|
-
key: colNode.getAttribute('data-col'),
|
|
840
|
-
label: colNode.get('innerHTML') || ' ',
|
|
841
|
-
sortable: !colNode.getAttribute('data-nosort'),
|
|
842
|
-
className: colNode.getAttribute('class'),
|
|
843
|
-
type: colNode.getAttribute('data-type'),
|
|
844
|
-
allowHTML: colNode.getAttribute('data-html') === 'true' || colNode.getAttribute('data-fmt') === 'html'
|
|
845
|
-
};
|
|
846
|
-
col.formatter = formatters[colNode.getAttribute('data-fmt')] || defaultFormatter;
|
|
847
|
-
cols.push(col);
|
|
848
|
-
});
|
|
849
|
-
return cols;
|
|
850
|
-
}
|
|
851
|
-
|
|
852
|
-
function getRowData(trNode, cols) {
|
|
853
|
-
var tdNodes = trNode.all('td'),
|
|
854
|
-
i,
|
|
855
|
-
row = { classes: {} },
|
|
856
|
-
node,
|
|
857
|
-
name;
|
|
858
|
-
for (i = 0; i < cols.length; i += 1) {
|
|
859
|
-
name = cols[i].key;
|
|
860
|
-
node = tdNodes.item(i);
|
|
861
|
-
row[name] = node.getAttribute('data-value') || node.get('innerHTML');
|
|
862
|
-
row[name + '_html'] = node.get('innerHTML');
|
|
863
|
-
row.classes[name] = node.getAttribute('class');
|
|
864
|
-
//Y.log('Name: ' + name + '; Value: ' + row[name]);
|
|
865
|
-
if (cols[i].type === 'number') { row[name] = row[name] * 1; }
|
|
866
|
-
}
|
|
867
|
-
//Y.log(row);
|
|
868
|
-
return row;
|
|
869
|
-
}
|
|
870
|
-
|
|
871
|
-
function getData(tbodyNode, cols) {
|
|
872
|
-
var data = [];
|
|
873
|
-
tbodyNode.all('tr').each(function (trNode) {
|
|
874
|
-
data.push(getRowData(trNode, cols));
|
|
875
|
-
});
|
|
876
|
-
return data;
|
|
877
|
-
}
|
|
878
|
-
|
|
879
|
-
function replaceTable(node) {
|
|
880
|
-
if (!node) { return; }
|
|
881
|
-
var cols = getColumns(node.one('thead')),
|
|
882
|
-
data = getData(node.one('tbody'), cols),
|
|
883
|
-
table,
|
|
884
|
-
parent = node.get('parentNode');
|
|
885
|
-
|
|
886
|
-
table = new Y.DataTable({
|
|
887
|
-
columns: cols,
|
|
888
|
-
data: data,
|
|
889
|
-
sortBy: 'file'
|
|
890
|
-
});
|
|
891
|
-
parent.set('innerHTML', '');
|
|
892
|
-
table.render(parent);
|
|
893
|
-
}
|
|
894
|
-
|
|
895
|
-
Y.on('domready', function () {
|
|
896
|
-
replaceTable(Y.one('div.coverage-summary table'));
|
|
897
|
-
if (typeof prettyPrint === 'function') {
|
|
898
|
-
prettyPrint();
|
|
899
|
-
}
|
|
900
|
-
});
|
|
901
|
-
});
|
|
902
|
-
</script>
|
|
903
|
-
</body>
|
|
904
|
-
</html>
|